Performance Parameter Estimation of Aircraft Auxiliary Power Unit Via A Fusion Model

Abstract

The Auxiliary Power Unit (APU) is designed to provide power and compressed air to the aircraft independently. By estimating the performance parameter of APU, its potential failure and abnormal information can be perceived in advance. To obtain accurate estimation result, Long Short-Term Memory (LSTM) network and Support Vector Regression (SVR) model are fused by Kalman Filter (KF). In this approach, LSTM network model is used as the state equation and SVR model is used as the observation equation. The effectiveness of this method is verified by adopting the real data of APU from the China Southern Airlines Company Limited Shenyang Maintenance Base.